Loading ToolBench
Agent context and repository readiness toolkit for AI-assisted development. Validates agent instructions, scans repositories, optimizes context and enforces evidence-backed workflows. Offline-first and deterministic.
ACKit provides 11 tools with consistent naming (verb_noun pattern: ackit_scan, ackit_doctor, etc.) and reasonable descriptions (avg ~80 chars). All tools are READ_ONLY, which is appropriate. However, parameter descriptions are sparse or missing entirely for several tools. Input schemas are present but minimal, most tools accept 0-2 params with basic type info. Output schemas are not documented in the visible code. Error handling guidance is absent. Tool composition is reasonable (each does one thing), but descriptions lack the depth needed for LLM selection in ambiguous scenarios.
Quick repository health summary (config + tasks integrity)
Return one intent document by id (normalized contract + fingerprint)
Return one task document by id
Return the resolved instruction graph JSON
List discovered agent skills
List docs-first tasks
Build a budgeted deterministic context pack with safety gates
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields without knowing what each tool returns. ackit_scan, ackit_doctor, ackit_instruction_graph, ackit_list_skills, ackit_validate_skills, ackit_policy_check all lack return type documentation.
Parameter descriptions missing or minimal for tools with no inputs (ackit_doctor, ackit_instruction_graph, ackit_list_skills, ackit_validate_skills, ackit_policy_check). While empty input is valid, descriptions of what the tool returns and when to call it are absent.
No error handling guidance. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next. E.g., ackit_get_task and ackit_get_intent do not explain what happens if the ID is not found.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | 2026-07-28+ | v2 |
Resolve the effective offline policy (chain + digest)
Run the ACKit scan pipeline and return the canonical JSON report
Validate agent skills and return tiered issues
Workflow state for a task (profile, stage, required artifacts)
Descriptions lack context for LLM selection. 'Return the resolved instruction graph JSON' does not explain when to call this vs ackit_list_skills or ackit_validate_skills. Descriptions should state WHAT, WHEN, and WHY.
Parameter 'format' in ackit_pack uses enum constraint but description does not explain the difference between markdown and json output or when to use each.